Breaking
93% fewer safety incidents at Kellanova $287K saved in 90 days at Church & Dwight AI cameras ride the forklift, not the ceiling 11% UPH lift across 40+ CJ Logistics sites Meet AiOn, the agent that watches the floor Enterprise-proven since 2017 93% fewer safety incidents at Kellanova $287K saved in 90 days at Church & Dwight AI cameras ride the forklift, not the ceiling 11% UPH lift across 40+ CJ Logistics sites Meet AiOn, the agent that watches the floor Enterprise-proven since 2017
Company Dossier · Logistics & AI

The company teaching warehouses to see.

OneTrack.AI bolts AI cameras onto forklifts to catch what the spreadsheet never does - how product actually moves, and how people actually get hurt.

OneTrack.AI logo
The logo sits in a warehouse the way a good foreman does - quiet, everywhere, and paying attention. OneTrack has been watching floors since 2017.
The Thesis

A warehouse management system tells you what should have happened. OneTrack.AI shows you what did.

And it turns out those are two very different datasets - which is, more or less, the entire business.

Here is a fact about warehouses that is either obvious or deeply strange, depending on how much time you have spent in one: the software running a modern distribution center knows, with great confidence, what is supposed to be happening. It knows that pallet 4471 was scanned at dock 3 at 9:04 a.m. and is now, in the eyes of the database, resting peacefully in aisle G. What the software does not know is that a forklift clipped a rack on the way there, that the driver took a shortcut through a pedestrian lane, or that the top case is crushed. The system tracks transactions. It does not watch.

OneTrack.AI is, at its core, a company built on that gap. Founded in 2017 by Marc Gyongyosi - who was, at the time, doing robotics work tied to BMW and noticing that production lines ran to sub-second precision while the warehouse next door still had people hunting for pallets with binoculars - the company sells a deceptively simple proposition. Put a camera where the work actually happens. Let a computer-vision model watch. Then tell the humans what it saw.

The clever part, and the part that makes this a real business rather than a demo, is where the camera goes. The obvious move in warehouse vision is to hang cameras from the ceiling, which gives you a lovely and largely useless view of the tops of people's heads. OneTrack started with drones, decided drones were a headache, and landed somewhere better: the forklift. The forklift goes everywhere the work goes. It is already moving. Bolt an AI vision sensor to it and you have, essentially, a roving eyewitness on every important machine in the building, capturing the route taken, the way product is handled, and the condition of the load when it leaves.

What you do with that footage is the whole game. OneTrack runs it through deep-learning models on low-cost edge hardware and turns it into three things a warehouse executive genuinely cares about: fewer accidents, more throughput, and cleaner shipments. The safety product flags risky behavior - a near-miss, a pedestrian in the wrong place - and fires a real-time alert plus video evidence, so a supervisor can coach the behavior instead of writing it up after the ambulance leaves. That distinction, prevention versus paperwork, is one the company returns to constantly, and it is a reasonable thing to be smug about.

The Receipts

Numbers the customers put their names on.

93%
Fewer safety incidents (Kellanova)
$287K
Saved in 90 days (Church & Dwight)
11%
UPH increase, 40+ sites (CJ Logistics)
70%
Fewer claims (GE Appliances)

Figures reported by OneTrack.AI and its customers. Treat vendor-reported results as approximate and directional.

You cannot coach a behavior you never saw. Video does not make the warehouse safer by magic - it makes the conversation possible. The OneTrack.AI operating premise

The productivity story is quieter but arguably more durable. OneTrack watches for gap time - the dead minutes where a forklift is idling, a task is stalled, an operator is waiting on something upstream - and the company's stated philosophy is to stack small wins rather than chase a moonshot. A 10 to 15 percent throughput bump does not make headlines, but across a 40-site network like CJ Logistics, an 11 percent units-per-hour increase is the kind of number that pays for a lot of cameras. The quality product, meanwhile, uses the same footage to catch over, short, and damaged shipments before the truck leaves the dock, which is a polite way of saying it prevents the argument that would otherwise happen three days later over the phone.

More recently the company has leaned into the word of the moment: agentic. Its platform now includes AiOn, pitched as an AI operations analyst that reasons over the floor data and the warehouse's existing systems rather than just displaying a dashboard for a human to squint at. This is the fashionable frontier of AI - agents that do things - and it is worth noting that OneTrack is pointing one at a warehouse floor rather than an inbox. Whether AiOn is a genuine autonomous analyst or a very good copilot is the sort of thing that depends heavily on where you draw the line, and reasonable people will draw it in different places. What is not in dispute is that OneTrack has a large, proprietary, and unusually physical dataset to point it at.

The customer roster is the strongest tell. This is not a company selling to a handful of curious pilots. Kellanova, Church & Dwight, Hain Celestial, Ryder, CJ Logistics, ID Logistics, The Wonderful Company, Bigelow Tea, GE Appliances, Holman Logistics - these are large, unglamorous, cost-obsessed operations, and they are exactly the buyers who do not renew software that does not work. That OneTrack pitches itself as live "in days" rather than quarters matters here too, because the enemy of enterprise warehouse software is usually not the technology but the eighteen-month integration nobody finishes.

Safety Incident Reduction

What "before and after" looks like.

Kellanova93%
ID Logistics84%
GE Appliances*70%
Kimberly-Clark80%+

*GE Appliances figure reflects a reported reduction in claims. All figures are vendor- and customer-reported reductions over stated deployment windows.

What It Actually Does

One camera. Five jobs.

Sensors

AI Vision Sensors

Low-cost AI cameras and edge sensors mounted on forklifts and around the floor, capturing routes, handling, and load condition in real time.

Safety

Incident Prevention

Detects risky behavior and near-misses, fires real-time alerts, and delivers video evidence for coaching - targeting 90%+ fewer incidents.

Productivity

Throughput

Finds gap time and idle hours, typically lifting throughput 10-15% by turning dead minutes into moved product.

Quality

Shipment Validation

Catches over, short, and damaged (OS&D) issues at the dock, before the truck leaves and the dispute begins.

Fleet

Asset Utilization

Monitors forklift-fleet usage to surface idle assets and improve how equipment gets deployed across a site.

Agent

AiOn

An agentic AI operations analyst that reasons over floor data and connected WMS/ERP systems - not just another dashboard.

Who's On The Platform

Big, unglamorous, cost-obsessed - the good kind of customer.

Kellanova Church & Dwight Hain Celestial Ryder CJ Logistics ID Logistics The Wonderful Company Bigelow Tea GE Appliances Holman Logistics States Logistics
The Path

From drone idea to deployments.

  • 2017

    The gap gets noticed

    Marc Gyongyosi, working on robotics tied to BMW, clocks the visibility gap between precision production lines and improvised warehouses.

  • 2018-19

    Cameras beat drones

    The drone concept gives way to AI cameras on forklifts. First commercial deployment lands in 2019.

  • 2020-24

    Enterprise roster grows

    Consumer-goods giants and major 3PLs adopt the platform, citing double-digit productivity and steep safety gains.

  • 2025-26

    Agentic turn

    OneTrack reframes around agentic AI, launching AiOn and "Agentic AI for Supply Chain."

How They Work

Five values, no fluff.

Sharp, Simple Solutions

Clear-cut approaches that repeat and scale.

Do The Work

Field execution over perfectionism.

Stack Small Wins

Compounding gains, not moonshots.

Humble Ownership · Continuous Reinvention

Outcomes over titles; challenge the assumptions as the market moves.

"OneTrack has helped customers like Kimberly-Clark and Bigelow Tea reduce safety incident rates by more than 80% and improve employee productivity by more than 30%." Marc Gyongyosi, Founder & CEO
Watch

Interviews & product demos.

OneTrack.AI on YouTube

Product walkthroughs, the AiOn agent, and warehouse footage from the company channel.

youtube.com/@OneTrack-AI →

The Warehouse of the Future

Founder Marc Gyongyosi on why vision beats transactions - podcast & interview series.

Listen / watch →
The Rolodex

Where to find OneTrack.AI.

Sources: OneTrack.AI, Forbes, Crunchbase, PitchBook, Tracxn, Inbound Logistics, States Logistics. Metrics are vendor- and customer-reported and should be read as approximate.